Premium
A new stable downward continuation of airborne magnetic data based on Wavelet deconvolution
Author(s) -
Abedi Maysam,
Gholami Ali,
Norouzi GholamHossain
Publication year - 2014
Publication title -
near surface geophysics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.639
H-Index - 39
eISSN - 1873-0604
pISSN - 1569-4445
DOI - 10.3997/1873-0604.2014027
Subject(s) - deconvolution , wavelet , tikhonov regularization , algorithm , gaussian , noise (video) , synthetic data , computer science , mathematical optimization , geology , mathematics , inverse problem , mathematical analysis , artificial intelligence , physics , image (mathematics) , quantum mechanics
This paper describes an efficient wavelet‐based deconvolution method for the downward continuation of airborne potential (magnetic and gravity) field data. We formulate the downward continuation process as a linear ill‐posed deconvolution problem. To obtain a reasonable downward continued field data, it is stabilized in a wavelet domain by minimizing the L1‐norm of the coefficients subject to the constraint that is the agreement of the upward response with the original observed data up to a white Gaussian noise level. The resulting convex constrained problem is then solved very fast and efficiently via the split Bregman iterations method. The generalized cross‐validation (GCV) criterion as a plotted curve versus the iteration count with new formulation is used to determine the optimum number of Bregman iterations without prior knowledge about noise properties. A synthetic magnetic field data embedded in a Gaussian noise is constructed from isolated multi‐source anomalies, whose results show that the airborne magnetic data can be continued stably downward by the proposed automatic sparse deconvolution method in a few iterations. When compared with conventional methods such as Tikhonov and Edge‐preserving with the proposed method, similar results were obtained. To test the performance on real data, we chose to use the airborne magnetic data of the central Iranian volcanic‐sedimentary belt. The enhanced downward continued data to a depth level of 200 m beneath the ground surface indicates adequate matching of high potential zones with previous working mines and copper deposits. Obtained results from both synthetic and real data confirm that the performance of the proposed technique is as good as the classical deconvolution ones.